AI Cybercrime Detection, Digital Forensics, and Governance: Scoping Review with ☸️SAIMSARA

To map and synthesize original studies addressing AI cybercrime, with emphasis on how AI is used to enable cybercrime, detect or investigate cybercrime, preserve digital evidence, support victims and law enforcement, and shape governance, adoption, and real-world risk. The review uses 162 references and builds its evidence map from 183 original studies with 61140123 total participants/sample observations (topic-deduplicated ΣN). This scoping review suggests that AI cybercrime research is dominated by a dual-use dynamic in which AI simultaneously accelerates offending and strengthens defense, with AI-powered attacks reportedly surging 238% while defensive systems achieve high study-level accuracies across phishing, fraud, and forensic tasks. The most recurrent signal indicates that AI is effective for automating detection and forensic triage, including reductions in manual investigation time and improved evidence handling through blockchain-backed integrity mechanisms. However, the evidence highlights that performance claims rest on non-comparable datasets and limited adversarial testing, so trustworthy deployment will require explainable, legally admissible, and adversarially robust systems integrated with human oversight. Practically, this supports prioritizing XAI-enabled forensic and fraud-detection tools embedded in governed workflows rather than autonomous adjudication. Future research should establish shared benchmarks and prospectively validate AI forensic outputs against courtroom admissibility and evolving offender adaptation.